ESXi host running on 6.0 may crash with a trace similar to :
0xnnnnnnnn :[ 0xnnnnnnnn ]Panic_Exception@vmkernel#nover+0x258
0xnnnnnnnn :[ 0xnnnnnnnn ]IDTReturnPrepare@vmkernel#nover+0x174
0xnnnnnnnn :[ 0xnnnnnnnn ]gate_entry_@vmkernel#nover+0x0
0xnnnnnnnn :[ 0xnnnnnnnn ]VmkDrv_WakeupAllWaiters@vmkernel#nover+0xc6
0xnnnnnnnn :[ 0xnnnnnnnn ]VmkTimerQueueWorldFunc@vmkernel#nover+0x21f
0xnnnnnnnn :[ 0xnnnnnnnn ]CpuSched_StartWorld@vmkernel#nover+0xa2
Example of a actual crash :




This issue occurs due to one of these reasons:
  • A slow memory leak in the VMKAPI-char-metadata heap.
  • When the ESXi host is using the mpt2sas or mptsas drivers.




This issue is resolved in VMware ESXi 6.0, Patch Release ESXi600-201608001. For more information, see VMware KB:2145663

To work around this issue:
  1. Take SSH session to the host.
  2. To check the heap memory usage, run this command:

    vsish -e get "/system/heaps/`vsish -e ls /system/heaps | grep VMKAPI-char-metadata`stats" | grep "percent free of max size:"
  3. If the heap memory usage is identified to be above 90%, schedule a planned reboot of the host.

TAGS : | | |